Discovery in complex adaptive systems

Regimes of collective logic

“…This procedure amounts to extracting one bit of information from 2,000 Â log 2 N bits encoding dynamics of the mean state parameter (2) which in turn are aggregated from 2,000 Â N bits describing the detailed state of the group during the decision making process in stage 2. This reduction exemplifies a coarse-graining approach in the study of complex systems (Flack, 2017;Spivey, 2018).…”

Temporary disruption in language processing reflected as multiscale temporal discoordination in a recurrent network

“…In this state-space reconstruction, the first time slice of an embedded 3dimensional trajectory carries with it information about the original time series' 1 st , 2 nd and 3 rd activity levels and the second time slice of this embedded trajectory carries information about the original data's 4 th , 5 th , and 6 th activity levels, etc. As long as the activity of the other 99 nodes is interdependent with the activity of the measured node, then embedding that one node's time series into multiple dimensions can produce a multi-dimensional trajectory that bears substantial resemblance to the overall network's own multi-dimensional trajectory (Spivey, 2018;Stephen, Boncoddo, Magnuson, & Dixon, 2009). Results show that a single bi-directional connection between the networks made it so that almost any randomly selected node from one of the original independent networks became correlated (or anti-correlated) with any randomly selected node from the other network (Falandays et al, 2020).…”

“…However, those frameworks sometimes leave unexamined the specifics of exactly how a reconfigurable network of cortical language subsystems (Chai, Mattar, Blank, Fedorenko & Bassett, 2016) achieves its smooth coordination during felicitous comprehension, and how it undergoes some discoordination during briefly infelicitous comprehension. By comparing the statistical character of a data stream that is extracted from a neural network simulation (where everything can be known about what's going on inside) to that of a data stream extracted from a person processing language input (where one must infer what is going on inside), we suggest that some progress can be made in understanding aspects and parameters of the simulation that might correspond to certain aspects and parameters in the person (e.g., Spivey, 2018). For example, there may turn out to be certain statistical characteristics in a time series of cognitive performance, such as multi-scale temporal structure (Van Orden, Holden & Turvey, 2003), that are naturally achievable with certain implementation-level models (Kello, 2013) but not naturally achievable with certain abstract computational models (Wagenmakers, Farrell, & Ratcliff, 2005).…”

Untitled

“…In this state-space reconstruction, the first time slice of an embedded 3-dimensional trajectory carries with it information about the original time series' 1 st , 2 nd and 3 rd activity levels and the second time slice of this … embedded trajectory carries information about the original data's 4 th , 5 th , and 6 th activity levels, etc. As long as the activity of the other 99 nodes is interdependent with the activity of the measured node, then embedding that one node's time series into multiple dimensions can produce a multi-dimensional trajectory that bears substantial resemblance to the overall network's own multi-dimensional trajectory (Spivey, 2018;Stephen et al, 2009). Results show that a single bi-directional connection between the networks made it so that almost any randomly selected node from one of the original independent networks became correlated (or anticorrelated) with any randomly selected node from the other network (Falandays et al, 2020).…”